Code Coverage

Code coverage is the fraction of a codebase executed while the tests run, reported per statement, per line, per function, and per branch. It is produced by Instrumentation.

What it can tell you is narrow and genuinely useful: which code no test has ever run. Uncovered code is untested by definition. That is worth knowing.

What it cannot tell you is whether the covered code is correct. A test that executes a function and asserts nothing produces identical coverage to one that specifies it exactly. This is why coverage makes such a poor target — see Goodhart's Law and Coverage Gate.

Two mechanical facts save a lot of confusion. A failing test yields no coverage number at all: the reporter aborts before writing its summary, so an unrelated red test produces an absent measurement, not a threshold failure. And percentages are the wrong unit for diagnosis; the underlying per-position records are what actually explain a shortfall, and a report format that emits them is usually one flag away.

Coverage also behaves oddly across multiple test environments measuring the same files, where a single logical position can end up recorded twice; see Branch Coverage.